By Laura Lardinois, PhD Candidate at McGill University

In the coldest months of the year, I had the chance to step briefly away from my thesis and immerse myself in records of summers spent in Akpatuurjuaq,ᐊ^ᑲᐸᑑ^ᕋᔦ^ᕋᑲ (Coats Island, Nunavut), much further north than I’ve ever been. This small island on the northern edge of the Hudson Bay has breeding colonies of thick-billed murres (Uria lomvia, akpa, ᐊ^ᑲᐸ), which, thanks to decades of annual surveys, are the most well-studied in the world. These Arctic seabirds, which are in the same family as penguins and puffins, are culturally significant for Indigenous communities and serve as indicators of environmental change.

Since the 1980s, teams of researchers from Environment and Climate Change Canada and McGill University have been going out to this remote island each summer to band birds, count population sizes, measure chick growth, observe feeding behaviour, and track reproductive success. This has generated huge quantities of valuable data, collected first on paper in the field, then entered into physical tables pre-Excel, then digitized in various spreadsheets and file types across the years. My job, as a data rescue intern with the Living Data Project, was to pick up where a previous intern left off a few years ago and wrangle these data into a clean, well-documented dataset that could be archived and made publicly available.

Field notebooks from the past few years (left) and paper logs (right) from the first 3 decades of monitoring (luckily, the data was already digitized for me)

At first, I was overwhelmed by the sheer number of files, not to mention the unfamiliar abbreviations, many tabs, and different information types. In some cases, having banded birds in a natural history course (thanks, Professor Trombulak!) came in handy when deciphering the data, but other information was more cryptic, especially without having been in the field to observe these murre colonies first-hand. Much like a novel starting in medias res, I was scrambling to get to know the cast of characters in these datasets, including not only the focal birds, but also the hundreds of observers who logged notes in field notebooks across four decades. Luckily, everyone who had been involved over the years was willing to help patch together their memories, including a memorable email sent from high elevations in the Himalayas to clarify chick hatching dates in the 1980s. I was also fortunate to visit the teams actively working on the project in person, spending a day at ECCC in Ottawa and another flipping through field notebooks and the original paper logs from 1981 through the early 2000s at McGill’s Mac Campus. 

Little by little, I began to latch onto patterns in the data and link them to conversations with the researchers who have led these monitoring efforts. For instance, some datasets had negative date values, which were the product of early years being recorded as “date since June 1^st”, since most of the breeding monitoring would start in June. However, in some years, crews came out early, resulting in -2 date values that needed to be converted back to May 29^th. Through the field notes, I caught glimpses of the reality of field work on Coats Island: sighting polar bears could end a day of monitoring, and several monitoring plots were relocated due to too much bear activity. Bird counts or other observations could be cut short by bad weather and poor visibility. 

Thick-billed murres, I learned as I tracked down hand-drawn maps of the study plots, nest on the very edge of rocky cliffs, effectively dotting the entire cliff face with black and white as they compete for prized nesting spots. Their eggs have a fun conical shape, which keeps them from rolling off the cliff. As I dug deeper into the data – and sent a slew of clarification emails to the amazing Coats Island research team – labels like “Q 37” changed from abstract alphanumeric codes to a specific nesting pair tucked into a crowded ledge.

Similarly, I learned to recognize the patterns of band IDs used to identify each bird: when caught, murres are banded with metal bands that have unique number codes to identify them when resighted or recaptured. In early years, bands had 8-digit codes starting with ‘996’, while recent years have switched to 9-digit codes. When they’re first banded, the complete band number is recorded, but sometimes, if you see the bird again through the binoculars, only parts of the band are visible, so I had to find ways to account for incomplete values. These band IDs are key, since they let us track an individual bird across time and link morphological data, its breeding site(s), potential partner(s), reproductive success, and any other data collected. 

With so many moving pieces, making lists of files, the date ranges they cover, what information they contain, and how they’re linked to other files, was critical to preserving my sanity. I also learned the value of making a changelog early on, to keep track of any issues I came across or modifications I made in a systematic manner. Usually, catching and trying to fix one issue in the dataset would dig up several others in the process, so being able to flag these to return to later was extremely helpful. Furthermore, tracking these changes systematically – alongside well-annotated code of what I’ve done – will allow others to pick up where I leave off and hopefully continue archiving new versions of this dataset after this summer’s field season.